GELLER Labs "Backyard Science"
Thoughts on a proton precession magnetometer design - a Proton Magnetometer Project
The goal of this project is a low cost high performance proton magnetometer (a digital magnetometer) kit for amateur scientists to be able to accurately measure and monitor changes in the Earth's total magnetic F field and to observe geomagnetic storms. There is a regular daily (diurnal) variation in the Earth's magnetic field. During events related to solar activity, there can be sudden changes in the field (such as a sudden impulse) as well as large excursions in the field which can be more than ten times the regular diurnal variation caused by magnetic storms.
(be sure to hit refresh to pick up our latest changes and entries)
Thursday, September 3, 2009
Well, it seems every couple to few years I start thinking about the famous C. L. Stong Scientific American Magazine, "The Amateur Scientist" column on Nicholas Wadsworth's proton magnetometer (more to follow on the Wadsworth article). Every time I go back and re-read the project, I marvel over how competent the description is and what a challenging design it must have been back then. Here is a low resolution snap-shot view of page 1 solely for purposes of identification; we are actively working to obtain copyright re-print rights. (Scientific American told us that they do not hold the copyright re-reprint rights for the column. Please contact us at the email address below if you are or know of the person or entity who is the present owner of these rights.) The full article can be still found in many public libraries or private collections. Also, the February, 1968 Scientific American issue occassionally shows up on eBay.
One part of his design that I am not sure about is his zener/diode combination that was supposed to shape the coil discharge curve. He says these components are "bridged across" his contacts, yet everytime I break out the colored pencils, it seems to me that the parts are in series (not across) and only for the "arc time" of breaking the circuit, then the cathode of the zener is hanging in free air (both of the contacts connected to the zener are open with no connection when the button is released). I wonder if this was an error in the design or in the schematic?
NOAA's Space Weather Prediction Center posts a webpage on Solar Cycle Progression. During this lull in solar activity before the solar cycle maximum, GELLER (Geller Labs) is working towards a low-cost proton magnetometer kit. The goal of our project is a low cost / relatively easy to build / high performance proton precession magnetometer suitable for home and school magnetic observatories. Probably (too early to know for certain) we will offer a box of parts including one or more printed circuit boards, most or all active and passive components, programmed micro(s) and/or the code and plans to build a stand and pick up coils (the sensor assembly).
Maybe it will connect to a PC program for the FFT (to read the frequency to a high enough resolution to see solar induced changes in the Earth's magnetic field) ... Back in 2003, I was using a Laptop microphone input for some experiments with the old Audigy (Cool Edit) software. Possibly this one could be a LabView 2009 stand-alone executable program made using National Instruments' application builder utility. If we go this route, the user would not need to own a copy of LabView.
I am experimenting this week with a simple dual relay circuit that adds a delay (the arm of the first relay) to allow for the coil energy dump into a resistor before connecting the coil to my resonating capacitor and amplifier. Eventually, the plan is to go to some sort of semiconductor switching, probably with FETs.
I made the first 125 ml bottle of distilled water by boiling doubly filtered domestic water onto a glass stew pot cover over the stove, then dripping it into a measuring cup. Wow, is that inefficient :). Since then I bought a small home distiller.
The signal levels are fine here, over 3V rms (the peak of the exponentially decaying sinusoid) in one test setup (Gain 41,000 (one of my custom amp prototypes, no where near a final design yet), followed by 50 (an old restored PARC 113 preamp), effective noise bandwidth 336 Hz, f0 2295 Hz for my location), however it is way too noisy here (1V) with a single bottle to do much signal processing. The noise floor was around 30 mV compared to the 1V noise pickup! A small magnet completely spoiled the decaying signal, demonstrating it was the precession signal and not just RLC ringing. (PIC1,(the small aluminum box is one of my old prototype multisage amplifiers), PIC2, PIC3, PIC4 (520 turns, #24 AWG wire, a "Wadsworth type" coil without his four separate windings to reduce self capacitance). Next job is to try the standard two bottle noise cancelling setup. Eventually, I hope to have the sensor, polarization switching, and front end amplifier outside, with a feed into the lab for "back end" development. The mosquitos are less bad this year, but they are out there!
Here in upstate NY -20F is not uncommon and -30F is not unheard of in the winter, so I guess I also need to experiment with fluids beyond distilled water. I am leaning in my present experiments towards a pair of 2" diameter PVC pipes with end caps (probably one rounded and one with the screw type insert; it's off to Lowes again tomorrow). It looks like my stock Nalgene plastic bottles will fit in the pipe. It's a little inefficient, since a small volume of plastic (the PVC wall and any space around the bottle, albeit tiny, is wasted by air), but should allow for easy testing of various types of fluids. Clearly some of these fluids are potentially dangerous as a fire hazard, so be warned, be safe and no, you cannot sue me if you burn your property :) On the other hand, the stuff seems safe on millions of shelves, so probably with just a little care, there is substantially little risk. Obviously, do not use a polarization current-time combination that gets the bottle too hot!
I am not sure yet if the PVC parts should only be dry fit, or if it makes sense to drill a hole or two as a pressure relief port? The plastic bottle top probably needs to be screwed down tight.
My understanding is that suitable proton rich fluids include kerosene, charcoal lighter fluid, and perhaps alcohol (hmm, I wonder about rubber cement?, some kinds of windshield washer fluids?). I have always been a little uncomfortable at the thought of large volumes of those sorts of fluids in a bottle or pipe with wires especially in these days where so many folks are scared of anything odd looking with wires, however a small 125 ml bottle (about 4 oz) should be completely safe and non-threatening. ahh, what times we live in.
Eventually I will post a reference page (like we did for the JCan project) of other known projects and writings.
Sunday, September 6, 2009
For noise cancellation I chose a non-conventional approach of connecting two side by side (both cap up) counter-wound (clockwise and counter-clockwise) Wadsworth bottles in parallel. (Note that turning a coil upside down does not yield a counter-wound coil. The counter-wound coil is literally wound in the opposite direction.) The bottles are separated by about 2" to suppress distortion of the standard external solenoid field between them (this distortion can be observed by looking at the inductance of the two side by side bottles at various separation spacing). I still have to work through the noise calculations, my reasoning was that since I already have enough signal from one Wadsworth bottle, any non-correlated noise (e.g. including thermal noise) would be reduced by sqrt 2 with the two bottles in parallel. Also, I did not want to raise the voltage needed for the polarizing current or double the inductance (which would extend the coil discharge time) as well as doubling the source impedance. Note that a series configuration (the conventional approach) tends to force you into a larger wire diameter (for lower resistance and to keep the required polarizing voltage down) which means higher weight and cost (i.e. more copper).
My "coil-connected" noise floor has dropped from over 1V to around 150 mV (net gain 2.05 million). Since one coil gives enough signal, only one side takes a fluid bottle with the other cancellation coil "dry". If I stay with this scheme, there is no reason to waste polarizing current in the dry side, so the polarizing switching circuit will take some more thought. Not needing to power both coils should give some nice power savings.
The latest coils (not necessarily anywhere near optimal) are 2" long wound with #24 wire 447 turns over a 4" long 2" diameter PVC pipe. (I was aiming for an even 1uf resonating capacitor, but missed and ended with another gaggle of parallel connected Cs. at .988 uf) Here is an online multilayer coil inductance calculator (no affiliation, works with IE, not FireFox). One end is a rounded cap that accepts the screw top of plastic Nalgene 125 ml sample bottle, the other end is a "clean-out" section with screw top. The clean out screw top goes on backwards so the square nub holds the bottle up near the coil. The edges of the cap and clean out parts act as walls for the coil. The same coil serves for polarization (on the order of 7 ohms, so less than one Amp at 6 V) and as sense coil. As set up now, only one side provides precession signal, while both sides provide ambient noise suppression (cancellation).
For best cancellation, this time, instead of scatter winding, I neatly wound tightly packed layers. Also, after the two coils were done, I matched their inductance (not the resistance) so that the transfer function of picked up ambient magnetic field to coil terminal voltage would match as well as possible. It seems just matching the number of turns is not good enough, since there can still be differences caused by the actual winding geometry.
Now that I have a reasonable signal to noise ratio, I looked at a few fluids in the 125 ml bottle (~4 oz). My home made distilled water gives the strongest signal (data to follow), and lasts the longest at about 2.5 seconds. Charcoal Lighter fluid gives less amplitude and the precession signal (2295 Hz here) lasts only about 1.5 seconds. Diesel fuel gives the lowest amplitude signal and and the signal lasts less than one second. All are "readable" near the peak, leaving open the question of whether to heat distilled water (during freezing weather) or to search out a fluid that will give a high enough amplitude signal that lasts long enough and that will not freeze down to about -30F. I still need to try out rubber cement, guess that bottle will be wasted if it does not work.
Now that the system is running, another experiment that I want to try is to run with tap water, filtered water, and the distilled water from the drug store that failed in previous test setups (perhaps for reasons other than water purity). My understanding is that tap water should not work ... we will see.
Also, now that the S/N is relatively good, I want to try some experiments with a triggered counter interval. Eventually I think a low cost experiment only makes sense with software signal processessing, however it would be interesting to see what reliable resolution can be achieved with various commercial counters and varying gate times within the first second or so of the precession signal.
Monday, September 7, 2009
The ambient noise at 2295 Hz with the coil connected is still well above the thermal noise of the coils (around 10 mV with my eff noise bw and gain). Also, the ambient coil connected value was up from around 0.12 V to 0.15 V to about 0.2 V today (following gain 2.05 million). I am thinking it might be worthwhile to think more about the coils and front end amplifier connections. I think the old prototype that I am using (I need to double check) is a direct connected single LT1028 unbalanced front end (I don't recall inverting or non-inverting; if I can't find the docs from last round, I will just "read" the board). One of my other prototypes was an instrumentation amplifier balanced front end, I think it was a TI microphone amplifier. There may also be common mode issues, possibly related to capacitive coupling to one or both coils. Also, even though I balanced the inductance, I am wondering if a trimmer in one or both sides might be helpful as a final fine tuning for cancellation. A common mode choke might be helpful too, need to look at where the roll off would be for my proto circuit and stock two winding ferrite cm chokes. Or, perhaps for a fixed observatory sensor, there could be additional small coils for fine tuning ambient field cancellation (seems too complicated and would be useless for a portable instrument, just a thought). Lastly, I should confirm that the spacing between the bottles does not have strong affect on cancellation. The argument (coil spacing) is probably too wide a spacing causes the two bottles to see different ambient fields and too narrow a spacing (side touching side) causes their solenoid fields to interact (reducing the inductance of both coils).
Tuesday, September 8, 2009
Rubber Cement appears to be a resounding success, definitely the best fluid tested to date. I used "Elmers" 4 oz bottles (just over one to fill my 125 mL Nalgene bottle. It polarizes in less time than the distilled water and has a comparable or slightly larger peak amplitude. At times it appeared to give a wider signature on the FFT, but that might just be the FFT math and how the energy was divided between the frequency bins (on a LeCroy scope). I have not put much work into the FFT side of things yet. I wonder if it is worth trying various types of silicon RTVs?
No thoughts yet on how rubber cement might do over temperature or how Rubber cement wet fluid (today's work) compares to when it dries solid?
Moving the car from out in the driveway (about 73' from my backyard table) to in the garage (about 33' from my backyard table) gives a good 0.5 Hz swing in the frequency, 2295 Hz to 2295.5 Hz). I probably should set up a fluxgate probe near the bottle and see if the peak "F" vector changes measured by the fluxgate match the field change using frequency and the Larmor constant; a project for another day.
I did not get to do much work today on coil ambient pickup, except at the end of today's work, I cut the ties hold the 2" spacer between my coils and moved them together and apart. As long as they were relatively parallel to each other, the ambient 180 to 200 mV today did not change significantly. Moving either of them off of parallel dramatically increases the terminal voltage, so cancellation is definitely working.
Saturday, September 26, 2009
I need to post more test results from earlier in the month. The rough saw horse testing continues ... pic1, pic2, pic3, pic4, pic5. I picked up an old Dell Latitude C640 (surprisingly robust for an old P4!) to run an old GPIB PCMCIA card in the field (here for dumping LeCroy Scope data, later for use with LabView in the field.)
In summary, I am still thinking about coil cancellation of ambient noise and why my ambient pickup is still relatively high.
I will begin to post some references that I am finding on the web. I came upon an interesting web page today on a PPM construction book, Signals from the Subatomic World: How to Build a Proton Precession Magnetometer. I had seen this model some years back as an assembled physics demonstration project by the same company, Extrom, LLC. As in any such project there is generally no right or wrong design. With this sort of project, a reliable working model is remarkable in itself. As I comment on other projects I come across, no negative criticism is intended, only my thoughts on where my project is going. I ordered the Hollo's book from Amazon. The Magnum experiment was originally a several thousand dollar purchase, but no criticism there, as we at Geller Labs are still operating at a severe financial loss. Also, many thanks to the Hollo (brothers?) for publishing their most interesting design as DIY project. The goal for the GELLER PPM amateur science project is under $250, possibly as low as $150 to $200 for parts alone (probably not including the copper wire). Probably the plan is to use some sort of PC interface, although a micro board is still a viable option.
One thing that strikes me about the Magnum is the ability to work at the end of a relatively long twisted shielded cable by use of a high quality instrumentation amplifier front end with both counterwound side by side sensor coils using a common ground connected to the cable shield. This technique is well known, never-the-less, it is still a feat to preserve the tiny precession signal to the end of a 50 foot cable.
They use a common larger diameter wire solenoid surrounding both counter-wound sensor coils for polarization. One advantage is the ability to use a larger diameter wire solely for polarization and to simplify the design by separating the polarization and sensing functions. A disadvantage is that it needs more copper and the common direction polarization field means they can only put a sample in one side (two sides cancel). My current understanding is that when using counter-wound coils for both polarization and sensing, the polarization field is reversed as appropriate, so both sense bottles can be used. Although, recall from early experiments above, one bottle gives plenty enough signal.
I am still leaning towards a version where two counter-wound sense coils are used, however only one of the sense coils contains a bottle and is polarized using the same sense coil for double duty (polarize-sense). In other words, there are two identical coils for sensing, but only one is powered during polarization and the other is out of the circuit during polarization (there is no point in dumping energy into the coil without a sample).
Questions being pondered: Should the polarization power be supplied by a small lead acid battery out near the sensors so as to eliminate the "heavy" copper polarization cable? If so, should it be solar cell charged? Should the front-end amplifier be out near the sensor, so signals sent back to the lab for post processing (filtering, fft, etc) are higher level, or digital? Should the raw pre-amplified or digitized signals be sent back by radio (Wifi, Zigbee, etc)?
I would like to get a sensor running in the field before the end of fall, even it just runs on staggered relays for now, to have the comfort of thinking about signal post-processing in the warm and comfortable laboratory.
QUESTIONS/COMMENTS/notice of typos, etc. send email to joegeller @ gellerlabs dot com
COPYRIGHT © 2009 JOSEPH M. GELLER, All rights reserved.